| matmarex | |
| May 13 2022, 10:17 PM |
| F35139299: image.png | |
| May 13 2022, 10:17 PM |
Something very weird has happened at https://gerrit.wikimedia.org/r/c/mediawiki/extensions/VisualEditor/+/763604:
I think its just some funky display issue, as the comment is properly attributed to you if we look at the API
{ "author": { "_account_id": 417, "name": "Bartosz Dziewoński", "email": "matma.rex@gmail.com", "username": "matmarex" }, "change_message_id": "c11e04032c1f88d6077c29d950f2186dd36156b3", "unresolved": true, "context_lines": [], "source_content_type": "", "patch_set": 9, "id": "39dd4bf5_919e942d", "in_reply_to": "e1c2630a_304ad77e", "updated": "2022-05-13 22:11:47.000000000", "message": "We should probably fall back to reloading the whole page in this scenario, we do that on desktop: https://github.com/wikimedia/mediawiki-extensions-VisualEditor/blob/6b6b065e073d59380e4ff2384d8d43a05460fbb6/modules/ve-mw/init/targets/ve.init.mw.DesktopArticleTarget.js#L667-L669", "commit_id": "e532af95c9c77365aa1c7ce543467e3ab7c38494" }
How much do you wanna bet that "updated": "2022-05-13 22:11:47.000000000", is gonna be real close to jenkins-bots'?
The whole history of comments is stored directly in git and can be retrieved and browsed using:
cd mediawiki/extensions/VisualEditor git fetch origin refs/changes/04/763604/meta git log -p FETCH_HEAD
Or that can be done via the Change REST API.
The jenkins-bot message on patchset 11 happened at Fri May 13 22:11:47 2022 +0000 and Bartosz message happened within the same second Fri May 13 22:11:47 2022 +0000. It is thus an UI glitch and @TheresNoTime won the bet :]
Maybe it has been addressed in a future Gerrit version (we run 3.4.4, they have released 3.5.x 3.6.x).
My guess is the issue comes from polygerrit-ui/app/elements/change/gr-messages-list/gr-messages-list.ts probably somewhere in _computeCombinedMessages which seems to compare date strictly with < and does not take in account it can come from different persons:
if (rDate < mDate) {
combinedMessages.push(reviewerUpdates[ri++]);
rDate = null;
} else {
combinedMessages.push(messages[mi++]);
mDate = null;
}I guess it can be reported to upstream